AI could take over most coding tasks within a year, Anthropic CEO warns
Speaking at the World Economic Forum in Davos, Anthropic chief executive Dario Amodei said rapid advances in artificial intelligence could soon automate much of the work traditionally done by software engineers.
ADVERTISEMENT
Artificial intelligence could dramatically reshape software engineering far sooner than many expect, according to Anthropic CEO Dario Amodei, who said most coding tasks may be handled by AI within the next six to twelve months.
Amodei made the remarks during a discussion at the World Economic Forum in Davos focused on how AI is transforming work and society. He said the shift is already underway inside Anthropic, where engineers increasingly rely on AI systems to generate code, leaving humans to review, refine and guide the output rather than writing software line by line.
According to Amodei, this workflow has already improved productivity and reduced development time. He suggested the next phase, where AI manages the entire coding process end to end, could arrive quickly, given how fast the underlying models are improving.
Also read: Davos 2026: Why Google and Anthropic are resisting ads in AI chatbots as OpenAI experiments
He said AI systems are rapidly approaching the point where they can perform nearly all the core functions of a software engineer, from writing and debugging code to optimising and maintaining it. While predicting exact timelines remains difficult, Amodei noted that progress continues to outpace expectations, even among those building the technology.
The comments have intensified concerns about job security in the tech sector. Software development has long been viewed as a resilient career path, but AI-powered tools are lowering the barriers to building applications. New platforms now allow users to describe requirements in plain language, with AI handling the technical implementation behind the scenes.
Amodei also pointed to a feedback loop accelerating AI development itself. As models become better at writing code, they help speed up research and product development, further shortening innovation cycles. Hardware constraints, such as chip manufacturing, may still slow certain aspects of progress, he said, but they are no longer the main limiting factor.
The Davos discussion included other prominent industry leaders, including Google DeepMind CEO Demis Hassabis, and explored how AI could affect employment across multiple sectors, not just technology.
Anthropic’s own AI system, Claude, has been central to the company’s internal shift. Over the past year, the model has expanded from conversational tasks into areas such as coding assistance, testing, editing and basic computer operations, highlighting how quickly AI capabilities are broadening.
While some experts argue that AI will primarily act as a productivity tool for engineers, others warn that junior and entry-level roles could face the earliest disruption. Amodei’s remarks suggest that the impact on software jobs may arrive much sooner, and be more far-reaching, than previously assumed.
